arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2103.02185cs.CVcs.LG
收藏

面向零样本学习的任务对齐生成元学习

Task Aligned Generative Meta-learning for Zero-shot Learning

University of New South Wales(新南威尔士大学);University of Technology Sydney(悉尼科技大学)

Zhe Liu, Yun Li, Lina Yao, Xianzhi Wang, Guodong Long

摘要:

零样本学习(Zero-shot learning,缩写为ZRL)是指学习从训练集(seen)中不存在的新类(seen)中分类实例的问题。大多数的分类器识别方法都是通过推断视觉特征和属性之间的相关性来训练分类器识别不可见的类别。然而,这样的模型在训练期间可能对所见的类具有强烈的偏向。已经引入元学习来减轻基础,但是当用于训练的任务从不同的分布中采样时,元学习方法是不适用的。在这方面,我们提出了一种新的任务对齐生成元学习模型的零样本学习(TGMZ)。TGMZ减轻了潜在的偏差训练,并使元数据库能够适应包含不同分布的真实世界数据集。TGMZ结合了一个属性条件的任务分布对齐网络,该网络将任务投影到统一的分布中,以提供无偏的模型。与最先进的算法相比,TGMZ在AWA 1、AWA 2、CUB和aPY数据集上分别提高了2.1%、3.0%、2.5%和7.6%。TGMZ在广义零样本学习(GML)设置中的表现也优于竞争对手3.6%,在我们提出的融合GML设置中的表现也优于竞争对手7.9%。

英文摘要:

Zero-shot learning (ZSL) refers to the problem of learning to classify instances from the novel classes (unseen) that are absent in the training set (seen). Most ZSL methods infer the correlation between visual features and attributes to train the classifier for unseen classes. However, such models may have a strong bias towards seen classes during training. Meta-learning has been introduced to mitigate the basis, but meta-ZSL methods are inapplicable when tasks used for training are sampled from diverse distributions. In this regard, we propose a novel Task-aligned Generative Meta-learning model for Zero-shot learning (TGMZ). TGMZ mitigates the potentially biased training and enables meta-ZSL to accommodate real-world datasets containing diverse distributions. TGMZ incorporates an attribute-conditioned task-wise distribution alignment network that projects tasks into a unified distribution to deliver an unbiased model. Our comparisons with state-of-the-art algorithms show the improvements of 2.1%, 3.0%, 2.5%, and 7.6% achieved by TGMZ on AWA1, AWA2, CUB, and aPY datasets, respectively. TGMZ also outperforms competitors by 3.6% in generalized zero-shot learning (GZSL) setting and 7.9% in our proposed fusion-ZSL setting.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑